Problem-Solving, Critical Thinking – and AI

Recently, I delivered a lecture for a banking audience on developing critical thinking and problem-solving – tailored to the age of AI.

The frame I use is a simple three-stage model of problem-solving I call QDC: Asking Questions, Divergent Thinking, Convergent Thinking. Each step asks something different of your brain, and each has a specific implication for how you use AI.

Q: Ask questions

Solving problems well starts with questioning them properly. Most of us do the opposite. We meet a problem with a quick brainstorm and immediate solutions, which feels productive but usually means we solve the wrong problem.

Try this now: Choose a current problem or challenge. Professional or personal, it doesn’t matter: how to overcome, improve, get better at, make more time for…

Then spend five minutes only asking questions about it. Write them all down, and answer none of them.

Some questions to get you going:

• How do you know it is a problem?
• What’s stopping you from solving it?
• What will define success?
• What will be the benefits?
• How much time will you give this?
• What advice would you give someone else with the same problem?

As you ask these, you will notice your perception of the problem changes. Someone who starts with “I can’t switch off” often ends with a different problem entirely, such as “I find it hard to say no”. That is what the reframe questions offer you.

With AI, this is hyper relevant. Judge a person by their questions rather than their answers, the saying goes. The AI equivalent: judge a person by their prompts, and you’ll know the quality of their answers.

Similarly, when AI hands you output, data, or information, you have to meet it with questions. What can be trusted here? What are the sources? What are the hidden incentives? 

D: Divergent thinking

Problem-solving then depends on idea generation: brainstorming options and potential solutions. AI can be genuinely useful here, but only if it is used well – and for this, the science is worth knowing.

Spend one minute coming up with different ways to use a paperclip, and notice what happens. Usually you’ll start with known uses and memory: ways you have actually used a paperclip – by using your default mode network and hippocampus.

Only when you run out of obvious, familiar answers do you shift into using your executive network and start genuinely engaging your mind, generating more useful and novel ideas.

The lesson is: for the best ideas, you need to fully engage your executive network, and one of the best ways to do this is shifting your perspective. Not using AI. I’ll get to that in a minute.

Working on your problem, give each of these three questions a few minutes:

1. Choose someone you respect. What would they tell you to do?
2. Imagine it is twelve months from now and the problem is solved. What did you do?
3. What could you do to make the problem worse? 

Notice the benefits of shifting your perspective, and how each question provides an entirely different lens through which to examine potential solutions. And now see what AI can offer – by offering multiple perspectives AND showing you the limitation and boundaries of your own perspective: your gaps, your biases, your blind spots.

C: Convergent thinking

Now you need to process and consolidate your ideas. Good decisions require you to extract meaning and insight. For this, the science is clear: you do not do it by working harder. Instead, you need to disconnect. As counterintuitive as it may feel, you need to stop working on the problem.

The reason you have flashes of insight or clarity in the shower is that you have shifted your brain state. The same reason you have good ideas on a walk, doing some light exercise, or as you are drifting off to sleep.

The three most reliable tools to achieve this are simple.

1. Disconnect and have a break.
2. Go for a walk or do some light exercise.
3. Sleep on the problem, or meditate

The relevance of AI here is again high, but for a different reason.

Quality thinking depends upon mental space and downtime. And now AI is making our sense of information and digital overload and time poverty even worse – and it’s fracturing our attention even more.

So as you use AI to solve problems, remember – quality thinking and judgement depend on the very thing modern work has made scarce. Disconnection and mental bandwidth. Protect them both fiercely.

Key points to remember

When faced with a problem:

1. Ask questions first. When using AI, put the work into your prompt, and meet the output with questions of your own.
2. Then shift perspectives. Generate solutions by changing your lens. Bring AI in only when your own thinking is done, to multiply your ideas and notice your blind spots.
3. Then disconnect. Give your brain the downtime it needs to process. Protect the walk, the break, and the night’s sleep as seriously as you protect the meeting.
4. Come back, decide and implement. This final step is when ideas turn into innovation.

I hope this has been thought-provoking. If so, please share it with someone else.

Phil